This is the second of two articles based on a conversation with Donald Allan, CFO of Fortune 500 toolmaker Stanley Black & Decker.
In the first part of our report, Don Allan talked about two types of transformation that he and his finance team are executing.
Both are technological in nature. One is a back-office reinvention that includes consolidating the company’s 70-plus ERP instances to fewer than 10. The other is what Allan calls “industry 4.0,” consisting of the manufacturer’s early efforts to create “smart factories” by leveraging digital technologies and robotics.
Here, Allan adds a third high-tech transformation to the list, and the one that might stretch manufacturers the most: increased usage of advanced analytics.
The changes you’ve told us about so far seem ambitious enough. Where do you stand with respect to using today’s most advanced analytical capabilities?
Most industrial companies, and we’re no different, haven’t dramatically evolved their business models to use advanced analytics to drive business models, business decisions, and even certain customer decisions.
Today we still do traditional data analysis. For example, in our business intelligence warehouse we have a lot of information on the profit margins of particular products. Why has this one gone down one or two points, and how can we begin to change that trajectory?
What’s changing now is that making decisions about pricing and manufacturing cost requires a lot more input than just the traditional internal data.
It requires looking at competitor pricing. It requires looking at competitors’ costs, where you can get those. And it requires looking at manufacturing-cost alternatives from vendors and things like that.
With that data, using artificial intelligence and machine learning algorithms, you [can develop] insights for evaluating your business model.
Are you using data scientists for that?
Yes, we’ve been bringing those folks on board. Sometimes I joke with the finance team that they’re all going to be replaced by data scientists.
But it’s only half joking. I really do think some of today’s finance functions are going to become data-science functions in the future.
So you need data scientists with a finance orientation — but hopefully you also have FP&A people with a data-science orientation.
Exactly right. Or, a handful of data scientists will do all the data crunching and provide insights, while the finance team figures out what to do with them.
We’re looking at ramping up the data-science team to about 75 people by the middle of next year. Right now it’s about 15 to 20 people. I’m not sure I want 75 people until they show me some value.
So we’re identifying use cases. [For each one, the data scientists] have to create value, whether it’s reducing inventory, creating revenue, lowering our costs, or whatever it might be. But I’m convinced we can create probably $200 million of value across the company over the next two to three years with this approach.
Can you give an example of a use case?
They’re trying to solve very common problems. For example, our tool business is very complex. There are too many SKUs. We’ve talked for years about how to reduce that complexity, focusing on SKUs that rotate very quickly.
It’s really going to be really interesting to see the outcome of the analysis on that.
But it’s one thing to get insights. Then you have to take action to drive value. If someone you bring in says you’re going to save $50 million by doing this or generate $100 million of revenue by doing that, it becomes hard to ignore that.
Given increasing reliance on analytics and robotics, is Six Sigma, which has been a staple process methodology in manufacturing for decades, destined to become passé?
Six Sigma uses mathematical techniques to prove out data before you make changes based on the data. That actually tends to slow down process change — it gets caught up in the mathematics.
But I think Lean Sigma, the streamlined version of Six Sigma, will always have value. It gets rid of the math and just looks at the process maps of how we do things today to find and eliminate waste points.
I think data and digital could replace that. But I’m not convinced yet. I don’t think Lean Sigma slows down what you’re trying to do.
Today, artificial intelligence could be considered the end game of advanced analytics. But it seems like every software company means something different when they tout their AI capabilities. What does the term mean to you?
To me it’s the same as machine learning. It just means learning from data and information. Based on data, information, and algorithms, a machine does something. Then it gets to a point where there is enough information and history that the machine makes changes to the algorithms automatically, without human intervention.
It can be as simple as processing things in a back office, telling you whether or not a bill should be paid or a journal entry should be recorded.
Do you worry that there might come a point where you’re losing control over what the machines are doing, such that you’re no longer able to see or understand the components of decision-making processes?
I don’t know. What we and a lot of other companies are doing is having control checks around the process. People are overseeing and reviewing the decisions being made.
I’m not sure that’s going to change dramatically. We’ll always want to monitor it. As the technology gets better and better, there might be a question as to whether the monitoring is part of the decision-making process or more of an after-the-fact review.
So, does it ever completely eliminate the human? I don’t think so. Well, I shouldn’t say that. Not in my lifetime, let’s say.
Where are you on the AI adoption curve?
Oh, not very far. If it were a baseball game, we’d be in the second inning.